Timeline for Linear model with categorical variables (and constant variable) in R
Current License: CC BY-SA 4.0
6 events
when toggle format | what | by | license | comment | |
---|---|---|---|---|---|
Aug 24, 2023 at 21:43 | vote | accept | Edge | ||
Aug 24, 2023 at 21:38 | comment | added | Peter Flom | It says that for each unit increase in age (probably years) there is a 0.00876 increase in LOS (presumably in days). So, assuming it is years and days, that's probably not important, even though it's significant. But it does change the other parameter estiamtes, which is what "confound" means. | |
Aug 24, 2023 at 21:13 | comment | added | Edge |
I was getting the 0.008762 from the estimate of age , wasn't sure what it specifically represented.
|
|
Aug 24, 2023 at 13:58 | comment | added | Peter Flom | Where are you getting 0.0087? It changes the intercept, and it changes scorefac2 from 0.43 to to 0.36 and scorefac3 from 0.53 to 0.43. Are those changes big? They seem pretty big to me, but it's a substantive question. You may also want to account for age because it's an important covariate. (I would think age is important in most studies related to health). | |
Aug 24, 2023 at 13:46 | comment | added | Edge |
Thank you. LOS is actually log(LOS) in this case. If I use scoreFac, am I right in thinking that lm(LOS ~ scoreFac + age) changes the estimate by 0.008762 (P=0.001), and thus it isn't a significant confounder (due to low change in estimate)?
|
|
Aug 24, 2023 at 13:28 | history | answered | Peter Flom | CC BY-SA 4.0 |